[lsb-bugs] [Bug 2581] xts5 tests Xlib12/sndevnt, Xlib12/staftrfnct fail on newer distributions

[bugzilla-daemon at linux-foundation.org]

http://bugs.linuxbase.org/show_bug.cgi?id=2581





--- Comment #3 from Stew Benedict <stewb at linux-foundation.org>  2009-05-06 09:58:13 ---
For sndevnt, it seems the "known bad event" type 35 is the problem. If I wrap
the test sequence in an idef to exclude this value, then it passes on
Ubuntu9.10, where it previously SIGSEGV'd. Here's the list of values actually
used after excluding 35:

520|0 21 00019615 1 12|TRACE: Using known bad event type 0.
520|0 21 00019615 1 15|TRACE: Using known bad event type 1.
520|0 21 00019615 1 19|TRACE: Using known bad event type 1.
520|0 21 00019615 1 22|TRACE: Using known bad event type -1.
520|0 21 00019615 1 25|TRACE: Using known bad event type 127.

Values are determined by:

/* Use notmember() to get some bad event types. */
        n = notmember(event_types, NELEM(event_types), &bad_event_types[2]);

/* Make sure this list also includes 0 and 1 which are specifically for errors
and replies. */          
        n += 2;                 

        bad_event_types[0] = 0;
        bad_event_types[1] = 1;

Test then iterates through this list of bad types and does:

/* Call XSendEvent to send bad event type. */
                event_send->type = bad_event_types[i];                          
                trace("Using known bad event type %d.", event_send->type);
                startcall(display);
                if (isdeleted())
                        return;
                return_value = XSendEvent(display, w, propagate, event_mask,
event_send);
                endcall(display);

And indeed, if we look at X.h on these newer systems we see:

#define GenericEvent            35
#define LASTEvent               36      /* must be bigger than any event # */

vs:

#define LASTEvent               35      /* must be bigger than any event # */

on the older systems. So 35 is no longer a "known bad" event, nor LASTEvent,
like we have it called out in LSB headers. Excluding this value from testing
seems reasonable.





-- 
Configure bugmail: http://bugs.linuxbase.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.